Entire server halts after sucessful end of job

[han810]

This is an intermittent problem, and I've had no luck solving it with tech support from CA, SONY or Dell. We have a PowerEdge 4400 with an Adaptec 29160, a SONY LIB162 (SDX700C) drive, running Windows 2000, Arcserve 2000

All have been updated to latest service packs, as instructed by the techs.

Here's what happens. After a sucessful completion of a job (backup, restore, merge - doesn't seem to matter) the ENTIRE SERVER HALTS. No keyboard or mouse response. All power lights are green on all hard drives, but absolutely no activity shows on any hard drive. Power cycling is the only thing that brings it back. This does not happen after every job, just often enough to cause problems.

There is nothing in the event logs, nothing in the Arcserve logs. I'm at a loss here....

 

[davidmichel]

Since it just halts like that forget about logging. Even if it was captured in a log because of the halt the log would not be saved and so the data would be lost.

For the system to hang like that it has to be something at a low level. Even a driver problem would result in a dump. My guess is something at the hardware level. Start by checking for the sharing of IRQs.

If that does not help then just move ARCserve to a different system. To me hardware is cheap and worth it rather than spending a lot of time troubleshooting out the cause.

 

[han810]

No point in moving it - the majority of the data is on this server - if I move it, the slow down will make the backup run too long - I'm already suffering slower throughput since upgrading to Brightstor from ARCServe 2000.

The server locked up again today, an I've replaced the controller and upgraded the firmware.

 

[davidmichel]

Did you check for the sharing of resources?
Use the lastest firmware throughout the system.
Is the hard disk and tape device on the same SCSI controller? If so add a second, one for each.

As for BEB being slower than ARCserve 2000, don't know what to say because all though there may be many with the same complaint I have not come across them. From all my testing and of those I've worked with BEB has been consistently better in throughput. Now I have come across some users who claimed it was slower but it did not stand up to examination. A common reason for example is a substantial amount of new data consisting of either compressed and or small files.

 

[han810]

Since the speed changed immediately after installation, I can safely say that it is the new software.

Would you clarify what you mean by shared resources? I saw no conflicts in IRQ, and it is set to NOT virus scan as it runs.

What log can I check?

 

[davidmichel]

Sharing of resources is two things using the same IRQ or memory address.

Check a log for what?
If for sharing of resources? then it is the SysInfo, Computer Management System Information.
If to troubleshoot the problem? None, as mentioned before since the system is locking up it is at a level lower than the OS.

Well you already replaced the controller, the mainboard would be a good next choice for replacement.

 

[han810]

New motherboard + cables installed, same old same old. I got a new version (11) from CA to install. I don't think that will work. The whole problem seems to somehow relate to the final SCSI command coming from Brightstor to the controller. I'm out of ideas. Guess I just live with it until the server gets replaced in June.

 

[intradev]

Han810 and others:

We have the PowerEdge 4400 with the 3QC and the 29160. We have an Exabyte 215M and BrightStor Arcserve 9. We recently upgraded from AS 2000, which made things MUCH slower and gave us many SCSI errors. We did all Dell, Adapetec, CA, and Exabyte firmware and driver updates (over 20). Did moving to Veritas do the trick? This setup has ALWAYS been troublesome but now it's unusable. Thank you in advance.